Laser desorption mass spectrometric studies of artists' organic pigments. - Chapter 3 An experimental strategy for LDMS investigations of paint materials and paint cross-sections

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

Laser desorption mass spectrometric studies of artists' organic pigments.

Wyplosz, N.

Publication date

2003

Link to publication

Citation for published version (APA):

Wyplosz, N. (2003). Laser desorption mass spectrometric studies of artists' organic pigments.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

Chapterr 3

Ann experimental strategy for LDMS

investigationss of paint materials and paint

cross-sections s

LDMSLDMS analyses of paint materials critically depend on the optimisation of thethe desorption and ionisation parameters. In this chapter we will discuss the effect ofof several experimental parameters, such as the laser power density, the repetition raterate of the laser shots, the ITMS mass cut-off and the CID conditions for MS/MS experiments,experiments, and determine their influence on the quality of the LDMS data. In additionaddition we will discuss significant issues such as the laser power threshold, the shot-to-shotshot-to-shot variation of the ion formation, the measurement of isotope patterns andand the rate of ablation. The aim of this study is to establish the optimal experimentalexperimental approach for spatially-resolved LDMS of artist's materials.

3.1.3.1. Introduction

AA great variety of experimental parameters influence the outcome of an LDMSS analysis. Formation of ions at the sample surface depends on the photon-samplee interactions, which is governed by sample arrangement, laser wavelength, laserr power density and repetition rate, surface temperature, chemical composition andd surface morphology. As LDMS aims at the determination of the local chemical compositionn of the surface, it is imperative to establish what experimental parameterss to use to answer a particular analytical issue. For instance questions suchh as what is the power density best used for the structural analysis of fragile organicc pigments in a complex matrix, what is the smallest pigment particle that cann be observed, etc, have to be addressed.

Inn this chapter we will discuss the method to obtain optimum LDMS conditionss for the investigation of paint materials and paint cross-sections. We will emphasisee the surface properties of the sample, the mounting of the sample in the MSS holder, the influence of the laser power density, the close repetition of laser shots,, the mass cut-off and collision-induced dissociation parameters of the ITMS analyser.. Examples of LDI and MALDI analyses of paint materials will be presentedd to illustrate these different issues. As a result, an optimal strategy to performm LDMS of paint materials and paint cross-sections is established.

3.2.3.2. Sample and sample mounting

3.2.1.3.2.1. Sample holders

AA large variety of sample types is investigated by LDMS in this thesis. Theyy include reference materials, calibrants, liquid oil pigment mixtures, cardboardd coated with one or several layers of pigmented medium, paint cross-sectionss and dyed fibres with threads of a few tens of micrometers in diameter. Thesee samples have to be mounted accurately in the focus of the desorption laser beamm in the ion source. Two sample holders compatible with both the TOF-MS andd the ITMS ionisation sources were designed to make LDMS possible with sampless of all different kinds.

Thee first and most common sample arrangement (Figure 3.1.A) is the use off a metallic substrate. A standard sample holder is used mainly for the analysis of

referencee materials with LDI and MALDI. Sample deposition is facilitated by the applicationn of a droplet of a solution or a suspension of the analyte under study, resultingg in a thin layer of analyte on the metallic surface of the probe after evaporationn of the solvent (Figure 3.2.A). Analyses of these thin layers are particularlyy successful in the case of organic pigments. The usage of thin films greatlyy enhances the signal in comparison with powder of coarser grain sizes, thickerr films, or paint samples of larger dimension.

Figuree 3.1 (A) Commercial LDMS probe surface with 26 pre-defined sample positionspositions for the study of thin films and absorbed particles. (B)

Home-buildHome-build LDMS probe for the study of paint cross-sections and paintpaint reconstructions on a cardboard support, front side. A

calibrantcalibrant has been deposited in the appropriate groove. (C) Back sideside showing the cavity where the sample is introduced and fastenedfastened with a back spring. (D) Home-build LDMS probe for the

studystudy of paint cross-sections and samples deposited on a TLC plate (E).(E). Probe holder with the XYZ manipulation system.

Ann alternate approach was developed to minimise the dispersion of the samplee at the surface of the probe. Small TLC plates (ca. 5x5x1.3mm) coated with cellulosee were mounted on a home-build sample holder as shown in Figure 3.1 D. Too facilitate the mounting of a TLC plate, the standard TOF-MS sample holder wass equipped with a 1.3 mm deep groove. TLC plates are fastened using double-sidedd sticky tape (Figure 3.2 B). A small aliquot of a sample solution is subsequentlyy deposited on the surface of the TLC plate. The higher surface tension off the cellulose surface ensures a minimum dispersion of the sample leaving a concentratedd spot of analyte. This preparation is especially useful when only small

amountss of material are available in dilute solution. The ionisation efficiency is similarr compared to the use of a metallic substrate. The TLC plates offer the added advantagee of providing a disposable substrate for the analyte, which eliminates the possibilityy of probe contamination from previous analysis.

Inn a similar fashion, cross-sections can be fastened in the groove of this sample-holderr using double-sided adhesive tape (Figure 3.2 C). This configuration necessitatess the correct trimming of the cross-section block of resin to fit the size andd depth of the groove. The main drawback of this method is that when the front andd back surface of the mounted samples are not parallel, differences in the z-axis positionn of the laser focus along the sample surface will occur. Larger samples suchh as embedded cross-sections than cannot be thinned down, irregular samples, paintedd cardboard and textile fibres require a different sample holder altogether.

Wee developed a sample holder for this purpose as shown in Figure 3.1.B andd C. The sample holder has a cylindrical cavity of 12 mm in diameter and 6 mm inn height with a circular aperture of 8 mm in diameter. A resin block, a piece of paintedd cardboard or a fibre thread is clamped inside the cavity with the aid of a backk spring (Figure 3.2 D and E). The spring-loaded clamping of the sample in the cavityy ensures that the sample surface is parallel to the surface of the probe holder andd limits the z-motion of the surface during sample positioning. A groove in the surfacee of the probe allows the deposition of a calibrant at the exact same level as thee sample surface. Home-made sample holders have the same dimensions as the regularr sample holder and are introduced in exactly the same fashion in the mass analyserr (Figure 3.1 E).

3.2.2.3.2.2. Level differences

Correctt positioning in the z-axis (axis of the extraction) of the samples in thee ionisation source of the TOF-MS is essential. Differences in ionisation positionss are expected to induce shifts in the flight-time that reduce the mass resolution.. Tests were designed and conducted to establish whether such a negative effectt could be observed within a range of a few millimetres. No significant shift wass observed for level differences within O.lmm-range. We explain this restricted effectt by the combination of delayed and pulsed extraction. For z-coordinates differencess in excess of 0.1mm a significant mass shift is observed. A shift of 2 Da forr 0.2mm and 10 Da for 1.3mm has been determined at a mass of 1000 Da. This hass dramatic effect in the analysis of textile fibres as will be discussed in more detaill in Chapter 6. The calibrant groove designed in the spring-loaded probe holderr ensures the same z-coordinate for the analyte and the calibrant.

Figuree 3.2 Different probe holders for spatially-resolved analysis of paint

materialsmaterials with TOF-MS and ITMS: (A) thin film deposited on a metallicmetallic probe (B) TLC plate and (C) embedded paint cross-section stuckstuck on the probe (D) embedded paint cross-section and (E) fibre clampedclamped in the holder's cavity.

3.3.3.3. Laser-sample interaction

Thee amount of energy deposited into the sample is a key to the understandingg of the interaction between the laser beam and the sample surface. In thiss section we will describe the parameters that determine how much laser energy iss converted into internal energy of surface molecules i.e., the amount of energy deposited. .

Energyy deposition at the surface of the sample depends both on macroscopicc properties and on molecular structure of the sample surface (chemical compositionn of the different compounds). The macroscopic properties include the compositionn and complex arrangement of the layers, which is expected to influencee the propagation of light at the surface of the section through different reflection,, scattering and refraction properties, and surface corrugation. The differentt molecular structures in complex samples lead to different molecular absorbancess at different wavelengths. For that reason a mixture of various paint materialss can hinder or inversely promote the LD1 process (acting as intrinsic matrix).. The type of medium, as well as the pigment-to-medium ratio in a coloured layerr will determine the energy transfer efficiency in the desorption process (from thee photons to the analyte molecules). High heterogeneity of the sample is expectedd to induce differences in photo-chemical and photo-physical processes fromm spot to spot at the surface of the sample, and possibly from shot to shot resultingg in laser ablation.

Thee amount of internal energy deposited and dissipated in the surface will affectt the ionisation efficiency, the degree of fragmentation, the desorption rate and thee overall interaction volume. Broadly speaking, two power density domains can bee distinguished "°. In the low power domain (< 108 W/cm") evaporation of surface layerss is obtained, partially in the form of intact neutral and ionised molecules. In thiss domain the evaporation process is rather reproducible and volume of material

ablatedd is kept small, typically in the order of urn3. There is usually no noticeable

damagee to the sample following a single pulse. Ionisation efficiency - i.e., the ratio off ions to neutrals - is low (of the order of 10"5) but the mild conditions assure the generationn of diagnostic ions from organic compounds up to a few 1000s Da. In higherr power domains, elemental ions and small molecular fragment ions are mainlyy desorbed, and ionisation yields are increased. The dimension of the ablation craterr scales with the power density. The analyte volume desorbed per laser shot (craterr volume) from a sample surface depends on a variety of parameters, the type off analyte, the surface condition of the sample and the characteristics of the laser beamm (laser-target angle, density, etc). In the low laser power domain, very small

ablationn volumes (<umJ) are not easily determined and the size of ablation craters iss rather estimated with the surface diameter of the impact. The size of the laser beamm does not provide a precise indication of the crater diameter since the desorptionn rate varies over the irradiated surface area. A rough estimation of the interactionn area is given by the observation of the fluorescence spot provided by thee sample under laser irradiation. A more accurate estimation is obtained by measuringg the diameter of a discoloured area left on a photosensitive paper, or the diameterr of a hole in a thin film of reference material (Figure 3.3). In spatially-resolvedd analysis, the size of the UV laser beam will be made as small as possible too increase the spatial resolution. The laser optics limits the beam diameter to 40umm (estimated with the discoloration of photo-sensitive paper).

Figuree 3.3 Crater left at the surface of photo-sensitive paper showing the size ofof the laser impact for a defocused experiment.

Ann experiment was performed where the laser energy of a UV laser beam workingg at 337nm was varied between 5 and 70 uJ under the same focussing conditionss (Figure 3.4). Tuning of the laser energy is obtained by attenuating the laserr beam with an optical filter positioned in its path. 0% attenuation corresponds too full transmission of the laser light whereas 50% attenuation means that approximatelyy half of the light emitted by the UV laser is filtered before reaching thee sample. This experiment demonstrated that the degree of discoloration varies moree than the size of the discoloured area. It is not possible to evaluate the real size off the impact when low laser energies were employed since the spots are not clearlyy defined. For that reason all spot sizes in this thesis are related to the spot sizee at maximum laser energy. This implies that the actual spatial-resolution is probablyy lower than the quoted spot size at low laser energies.

Figuree 3.4 Power density of the UV laser beam as a function of the energy

attenuationattenuation provided by the laser optics. A 0% attenuation correspondscorresponds to the full energy delivered by the laser; at 50% attenuationattenuation about 90% of the laser energy is filtered out by the attenuator. attenuator.

Inn a typical LDMS experiment, it is difficult to determine prior to analysis howw much energy per pulse should be applied to achieve precisely a given power densityy in the desorbed micro-volume. A strategy was devised in which the power densityy is progressively increased starting with a fully attenuated beam. After assessingg the LDI threshold, the laser energy is set to an optimum value as describedd in the next example. This example illustrates the relationship between thee laser power density and the spectral information for LDI of organic colouring compounds.. Kaempferol, a yellow organic colouring material belonging to the flavonoidss (see Chapter 4) was deposited as a thin film (approximately 10 urn) on aa stainless steel probe and analysed by TOF-MS. The 337nm output of a nitrogen laserr was focussed to the surface of the probe with a beam size of approximately 40nmm diameter.

Inn the lowest power density range (<10uJ), energy deposition is insufficient too induce desorption/ionisation process and spectra display nothing but the instrumentt noise. It is not impossible that the laser removes material from the surfacee of the sample, but this material is not detected by the mass spectrometer. Differentt reasons could account for this absence of detection: (1) material is desorbedd but not ionised, (2) ions are produced but do not reach the detector, (3) ionss reach the detector but in insufficient amounts, i.e. below the detection limit, or outsidee the analyser mass detection range (the TOF-MS has theoretically no mass boundaries,, but the ion injection conditions impose a low mass cut-off for the ITMS). .

Whenn a certain laser density is reached, called LDI threshold (ca. 1OuJ for kaempferoll on stainless steel), sufficient energy is deposited into the analyte to producee observable gaseous ions and the mass spectrum displays a series of kaempferoll peaks (Figure 3.5.A).

Att this laser density level, the smallest quantity of internal energy is impartedd to the molecules that lead to formation of intact molecular ions, i.e. the amountt of fragmentation is low under these conditions. During the LDI process, bothh positive and negative ions are generated but ions with only one polarity are acceleratedd towards the detector. Successive analyses of the same compound revealss that the minimum laser power density required to detect the presence of the moleculee (LDI threshold) depends on the arrangement of the sample film (thickness,, size of the aggregates) and varies from one desorption spot to another.

Figuree 3.5 Effect of laser density on the appearance of the LDI-TOFMS mass spectraspectra in the case of kaempferol. Spectra at the desorption and ionisationionisation threshold of 10 pJ (A); at higher density higher S/N is obtainedobtained and more fragmentation and dimers/clusters are observed (B),(B), at even higher density the detector saturates but no additional analyticalanalytical information is provided (C), highest laser densities lead toto excessive fragmentation and dimerisation, with saturation of the detectordetector (D).

Inversely,, different arrangements of the same compound produce different spectra (distributionn and relative intensities of fragment ions) for an identical laser power density.. For this reason, great care should be given to sample preparation and the exactt value of LDI threshold laser power densities cannot be proposed. Although similarr trends are observed comparisons between different components are difficult.. An experiment were different reference organic pigments were deposited ass thin films on stainless steel showed that under these condition the LDI threshold energyy fluctuates around lOuJ.

Thee threshold spec/rum of kaempferol in the positive mode (Figure 3.5.A) showss that low laser density LDI achieves soft ionisation. Absorption of the UV

laserr energy induces the formation of analyte radical cations M*+ and protonated

andd sodiated molecules [M+H]+ and [M+Na]+. Thermal ionisation of alkali

contaminantss - present in nearly all types of samples - generates alkali ions in high abundance.. The spectrum displays also fragment ions and a few ions of higher molecularr mass. Thermal energy imparted to the molecules during the LDI process iss moderate and pyrolysis (fragmentation through thermal activation) is greatly minimised.. Fragment ions with low relative abundances (0.5%) produce neverthelesss sufficient evidence to positively identify kaempferol.

Intens s 15000 0

10000 0

5000 0

2855 286 287 288 289 m/z

Figuree 3.6 LDI-TOF-MS of kaempferol. In inset, the theoretical isotopic

distribution. distribution.

AA closer look at the region around the molecular mass (Figure 3.6) shows

mostlyy protonated molecules [M+H]+ and only few radical cations M*+. Peaks at

2888 and 289 are attributed to the isotope 13C, which occurs in a natural abundance

off approximately 1.08 per cent*. Isotopic distribution for the 15 carbon atoms of

thee protonated ion of kaempferol of formula Ci5Hn06 gives relative intensities of

1000 for m/z M, 16.5 for m/z M+l, and 1.3 for m/z M+2. Isotopic distribution

Thee most common isotope of carbon is '2_{C, which nucleus has 6 protons and 6 neutrons, but about}

1,08%% of the natural carbon consist of the isotope 13_{C, which has one more neutron in its nucleus.} II i

providess essential information on the elemental composition and can be greatly contributee to the identification of the molecular structure 127

Itt is observed in Figure 3.5.B that an increase in the laser energy of 10% abovee the LDI threshold enhances directly the total number of ions collected by the detector.. At the same time neutral pressure is increased and we observe enhanced rearrangementt (cluster ions ) and oligomerization effects (dimers such as

[2M+H]+,, [2M+Na]+ are detected). At high laser power (Figure 3.5.C) mass

resolutionn gets worse as the peak width becomes broader. A large ion population resultss in a broader spatial distribution of the ion cloud. Higher ionisation yields promotee this broadening of the cloud through the so-called 'space-charge effect', thatt is an increase in repulsive forces between ions of the same charge. The detectorr saturates at approximately 30,000 counts, which makes measurements of relativee intensities above 30,000 counts impossible. At even higher laser energy densityy (ca. 70uJ) characterisation becomes even more difficult (Figure 3.5.D). Excessivee fragmentation is observed in combination with the formation of dimers andd higher oligomers (clustering). These clustering phenomena are attributed to the higherr neutral pressure in the laser plume at these high energies. The increased degreee of fragmentation is attributed to the higher amount of internal energy depositedd into the surface. Rising the laser power density results also in increased samplee consumption. High laser power density results eventually in serious damagee of the sample surface, which is visible under the microscope.

Inn conclusion, these experiments show that ionisation yields and structural informationn are interrelated parameters which depend on the laser power density. Additionall energy deposition amplifies the ion population (TIC) but promotes at thee same time the formation of a variety of additional ionic species (dimers, fragmentt ions). By tuning the laser power density correctly, it is possible to obtain ann optimal population of the characteristic ions. Low energy deposition during the desorption/ionisationn process increases the probability to obtain molecular structuree information by transferring intact molecular ions and diagnostic fragment ionss into the gas phase. Furthermore, it limits the sample consumption and preservess the integrity of the specimen surface for further analyses . Practice showss that sufficient variety of diagnostic ions with appropriate signal-to-noise ratioo is obtained for laser power densities within a small range (<5%) above the LDII threshold value. Working in this energy domain provides the optimal balance betweenn mass spectral information and sample consumption. Therefore this energy

'' A cluster ion consists of the main building blocks or structural moieties of the original molecule.

ff

High irradiance LDI could be interesting for speciation of inorganic substances. However, other analyticall techniques such as TOF-SIMS imaging (Heeren et al.) or SEM-EDX will be preferred in thiss case since they limit considerably the damage of the sample surface and provide a superior masss resolution.

domainn was consistently chosen for performance of LDMS experiments on paint materialss in this thesis.

3.4.3.4. Shot-to-shot variations

LDII experiments are usually performed with a series of several laser shots inn close succession, and the resulting spectral information is averaged to produce a reliablee mass spectrum. This practice is rationalised in the following experiment wheree we monitor the spectral information as a function of time during multiple-shott analyses. The flavonoid morin was deposited as a thin film on a stainless steel probee and analysed by LDI-ITMS. The UV laser was focussed to the surface of the probee and employed with a repetition rate of ca. 1 Hz. Phase-locking between the trapp function (i.e. ion accumulation and scanning) and the laser triggering assured thatt for each laser shot the spectrum is recorded under the same time conditions. Sincee repeated laser shots increase the sample consumption, the number of laser shotss (or duration of the LD process) should be theoretically kept as low as possiblee to prevent excessive damage of the sample surface.

Figuree 3.7 shows the total ion current (TIC) of morin, recorded for 45 secondss with an uncorrected offset of circa 2000 counts for the background. Duringg the time period t=[0-18s], the laser is still idle and the background is recorded.. At t=18sec the UV laser (355nm) is switched on and the total number of ionss detected increases shortly to an average value of 6000 counts. During the time periodd t=[18-33s], the TIC curve displays a serrated profile with shot-to-shot variationn in excess of 30% (i.e. three times larger that the background variation). In thee ensuing period t=[33-50s] shot-to-shot variations visibly subsides with values mostlyy under 10% (i.e. comparable to the background variation).

Intens. .

8000 0

6000 0

4000 0

2000 0

55 10 15 20 25 30 35 Time [s]

Figuree 3.7 Total Ion Current (TIC) of morin in multiple-shot LDI-ITMS

Laserr hardware is considered to have only a negligible effect on the variationn in TIC intensity since variation in power density from shot to shot should nott exceed 1%. The TIC profile is rather explained by modifications in time of the photon-solidd interaction as the distribution and relative intensities of analyte ions aree slightly different from shot to shot (no modifications were induced in the experimentall conditions during the measurement). Certain ions are only detected intermittently.. In Figure 3.8 the spectra for two successive laser shots are shown in parallel.. These irregularities are the direct consequence of shot-to-shot variations in energyy absorption. In most cases the fragmentation pattern of a molecule is manifoldd and consists of different fragmentation pathways. The quantity of internal energyy imparted to the molecules during the LDI process and subsequent collisions withh neutrals (neutral pressure, ion neutral ratio) determines the most energetically favourablee fragmentation pathways. Modification of the local surface configuration afterr each laser shot (ablation) is likely to slightly modify the conditions of energy absorptionn for the subsequent laser shots.

c c Z3 3 -Q Q < < 1008 0 --60 0 4 0 0 20 0 1008 0 6 0 --4 0 0 20 0 3 Q 3 3 2299 257 1 3 7 7

Ull.iii J UUuu i,iif.j ii. i.i| duA II .I'M.'

285 5 2 7 4 4 jl4Ji i _{M l l} 3 1 7 7 (Lull Ji i, uuu «HU U 2 2 9 9 303 3

L,il,ii rtlijt^Mili .[.i IM-PM

285 5 257, , UikJ J 274 4 U L L 3 1 1 JJk, , iJiiilliwii i.l i m i , 11 1255 150 175 200 225 250 275 300 325 Masss [u]

Figuree 3.8 Single ion current (SIC) ofmorin showing the shot to shot variation

Wee can clearly observe in Figure 3.7 that several laser shots (about five) aree necessary before a representative TIC signal is obtained, and a few more (about ten)) before this signal presents a steady profile. Several laser shots are clearly necessaryy to reach a steady state temperature at the surface of the sample. We proposee that the delay in appearance of ions is caused by a temperature dependence off the ion production. As the temperature increases, a plasma cloud is formed near thee surface of the sample. This idea has been proposed before to explain the formationn of sodiated and potassiated species under CO2 laser

desorption/ionisationn conditions H0. Additional absorption of the UV laser

radiationn in the plasma cloud in our experiments could increase the ion production byy ionising desorbed neutral molecules and by increasing the internal energy of the alreadyy existing ions. Temperature equilibrium is eventually reached.

Att low laser density the sample depletion is rather slow and several hundredss of shots are necessary to observe a meaningful decrease of the TIC. The cameraa output does not offer sufficient magnification to accurately monitor the modificationn of the sample surface during a multiple-shots experiment. However, sampless were examined after analysis under high magnification (x400) with an opticall microscope and it was not possible to observe any damage on the sample surfacee after ten shots.

Inn conclusion, LDI experiments should take into account the delay in ion productionn and the shot-to-shot variations by oversampling. Scans should be averagedd over several shots (typically about 10) to obtain representative mass spectra.. In the case of automated-scanning experiments, a minimum amount of shotss (between 5 and 10) should be used to reach the TIC steady state.

PaintPaint cross-sections

Thee complex micro-morphology of paint cross-sections and the limited amountss of analyte represents a particular inconvenience to use multiple laser shots.. Sectioned paint samples present a highly heterogeneous set of paint materials.. Different chemical compositions are found at different spots on the samplee surface, and deeper in the sample. Consequently, the composition of the analytee itself might change as the experiment goes along. In this case, recording spectraa by averaging ion signal on several shots (for a fixed location) is problematicall although it could contribute to statistically more relevant data. In generall ablation volumes should be kept as low as possible to limit sample consumptionn and safeguard the integrity of the layered structure. On the other hand,, if the compound of interest is not situated directly at the surface some etchingg of the surface layers is permissible.

3.5.3.5. TOF-MS versus ITMS: pressure and time-scale

Masss spectra obtained with the two mass analysers will be distinctively different,, even for identical compounds measured under similar LDI conditions (Figuree 3.9). Intens. . 80000 ' 60000 " 40000 " 20000 "

A A

L L

B B

Figuree 3.9 LDI of kaempferol at low laser power. The TOF spectrum (A) and thethe ITMS spectrum (B) that demonstrates the formation of cluster

Thee two analysers differ in two important aspects. First, the analysers' pressuress differ 5 orders of magnitude between both analysers. Secondly, the total analysiss time varies between microseconds for the TOF-MS to seconds for the ITMS.. These two differences lead to a different appearance of the analyte ions in thee different mass spectra.

Pressuree is measured at 10" mbar in the ITMS analyser compared to 10"7 mbarr in the TOF-MS. High pressure in the analyser of the ITMS will lead to an increasedd collision rate of the trapped ions with the helium background gas. These collisionss can have two effects, they can increase the amount of internal energy of thee trapped ions leading to an increased degree of fragmentation. On the other

hand,, the increased pressure can lead to the formation of cluster ions. In our experimentss we observe that the increased pressure mainly leads to an increased relativerelative amount of clusters in the ITMS under similar LDI conditions.

Thee period of time between ion formation and detection is significantly differentt in the TOF-MS and the ITMS experiments. In both cases we believe that differencee in time definition of the initially-generated ion packet only plays a marginall role since extraction is delayed and pulsed. Transport time to the detector iss estimated to be in the order of a microsecond in the TOF-MS compared to a few microsecondss in the low-energy transport system of the ITMS (note that ions producedd in the source of the spectrometer still have to be transported to the detector).detector). The total analysis time in the TOF-MS is solely governed by this transportt time. In the case of the ITMS, an additional accumulation, equilibration

andd detection time of ca. Is is required and determines the total analysis time. Duringg this extended time period, ions in the gas phase are likely to undergo furtherr transformations such as fragmentation through collisions with neutrals and formationn of cluster ions, or can be simply lost. This implies that the ion trap will detectt the metastable ions formed in the LDI experiment that are not observed in thee TOF-MS experiment.

3.6.3.6. Ion collection in the ITMS analyser: LMCO

Efficiencyy of the ion collection in the ITMS analyser must be optimised accordingg to the type of analyte and the type of MS experiments. The different parameterss that affect the ion collection efficiency are RF potential during ion accumulation,, the ion kinetic energy and the pressure in the ion trap.

Thee trapping RF potential of the ITMS analyser is a periodical wave of lMhz.. Ions can be accumulate in the ion trap only for a certain amplitude range of thee RF signal. In Chapter 2 we have discussed the connection between the frequencyy and amplitude of the RF field during ion injection and the lower mass limitt of trapping. The time domain acceptance of the ITMS can therefore be

improvedd by modification of the low mass cut-off Mcut.0fT l0°. This is illustrated

withh the following MALDI-ITMS analysis of polyethylene glycol (PEG)

HO[CH2CH20]nH,, the reference compound used as calibrant in our experiments.

Thee PEG sample from Serva (Heidelberg) presents an average molecular weight of 4000 Da. The matrix in the MALDI experiment was 2,5-dihydroxybenzoic acid (DHB)(DHB) from Sigma, Inc. The polymer sample was prepared by mixing a 1M matrix solutionn in ethanol with an approximately lOg/L analyte solution in ethanol, yieldingg a molar ratio matrix/analyte of 1000/1. The analyte and the matrix were depositedd as a thin film on a stainless steel probe. The UV laser was focussed to

thee surface for LD and analyses were performed with the ITMS. In Figure 3.10 we cann observe that MALDI spectra recorded for different values of Mcut-off (70u and lOOu)) give a different peak distribution for the same PEG sample. This is evidence thatt trapping efficiency depends on Mcut-off. In practice it implies that optimisation off MCut-off is required, especially when the highest amount of one particular ion is desiredd to perform multiple-stage MS experiments.

40 0 30 0 20 0 10 0 JJ 100 80 0 600 1 40 0 20 0 Mcu,.offf = 70Da 2733 I

II 1

B B

iwJiii1.ü1ii.Hiiiiiiiii.i»iiiinUffit^>iii<4*rt«kaJL^li>i)^iiill./iiiiiiiii i

1500 200 250 300 350 400 450

mkiêi mkiêi Masss [u]

Figuree 3.10 Collection efficiency of the ion trap: MALDI-ITMS of polyethylene glycolglycol (PEG), MWav=400, measured with Mcut-off of 70 Da (A) and

100100 Da (B) showing the dissimilar distribution of the ions collected.

Thee trapping characteristics also depend on the size of the ion cloud which iss governed by the kinetic energy spread upon generation of the ions. Trapping efficiencyy will vary for the different part of the ion cloud (0 to 100%), and significantt amounts of ions will be lost (50%) during collection. Narrow ion clouds (timee distribution < l u s ) are problematic since efficiency of the trapping depends stronglyy on the value of the RF amplitude when the cloud of ions reaches the entrancee of the trap (in the worse case, no ions are trapped). Fortunately, the low energyy ion transport system results in a kinetic broadening of the ion cloud. The laserr is fixed in synchrony with the oscillation of the RF signal and it has been shownn that the exact position of the pulse in the RF period does not change the reproducibilityy of the analysis. This indicates that the ion cloud must be wider (in time)) than a number of RF cycles.

Finally,, we should mention that correct trapping is only obtained for a limitedd ion population in the trap. For high quantities of ions (as discussed in previouss paragraph) it is recommended to limit the ion introduction time, i.e. the

timee period during which ions are allowed to enter the trap. This is achieved by applyingg a block voltage to the ion introduction lens. An excessive ion population inn the trap induces a space-charge effect that modifies the trapping potential and resultss in a deterioration of performance of the ion trap.

3.7.3.7. CID experiments with the ITMS analyser

Ionss trapped in the ITMS cell can be investigated in multiple-stage MS experimentss (MS"). Manipulation of the ions in the trap involves successive operationss in time (isolation, excitation, detection) that can be modified in order to optimisee the structural clues for molecular elucidation. In the illustrative MS/MS analysiss presented here, molecular ions of the flavonoid compound kaempferol weree investigated in a non-spatially resolved experiment. The laser was tuned at or slightlyy above the desorption-ionisation threshold where a continuous signal for the intactt molecular ion is recorded (Figure 3.1 LA).

Figuree 3.11 LDI-ITMS of kaempferol at the laser threshold: MS analysis (A)

isolationisolation of the pseudo-molecular ion m/z 287 at Mcu,.0ff= 125 Da (B)(B) demonstrating a clear increase of the signal to noise ratio.

Whenn the sample is in excess the laser power density can be further increasedd in order to enlarge the precursor ions population. Slightly higher irradiancee promotes the production of intact parent ions without excessive increase

off the fragmentation. A Mcu,.0frof 75 Da was chosen to obtain efficient trapping in

thee mass range of kaempferol (m/z 286). Figure 3.1 l.B shows the isolation step of thee molecular ions. Isolation spectra are habitually recorded prior to CID experimentss in order to optimise the population of the precursor ions. In this mode, alll ions except the molecular ion of kaempferol are expelled from the trap. In practice,, ions isolated in the trap are typically selected on a mass interval with a Gaussiann distribution of A(m/z) = 10 Da. Therefore, after isolation the spectrum stilll shows the isotopic distribution of the radical cation and protonated molecular

ionn of kaempferol on the interval [286-289]: [M*+] and [M+H]+ and their

respectivee isotopes. After isolation a new Mcut-off can be selected that best matches

thee next MS operation (for the purpose of this example Mcut.0ff = 125u). This

MS/MSS cut-off mass and the selected ion mass determine the RF characteristics (frequency,, amplitude) of the excitation pulse used for CID. The S/N ratio is generallyy significantly increased after isolation. The transmission of the ITMS duringg isolation is theoretically integral, i.e. all ions are retained for subsequent manipulationn which means an identical absolute intensity of the signal before and afterr isolation.

Alll MS/MS experiments in this thesis were performed with isolation of the precursorr ions prior to fragmentation. This operation greatly simplifies the interpretationn of the spectra since the relation between parent and daughter ions is unequivocal.. However, in a CID experiment it would be fully possible in theory to fragmentt one or several specific ions without proceeding to a prior isolation.

Thee isolated protonated molecular ion of kaempferol (m/z 287) was fragmentedd using collisional induced dissociation (CID) in the trap cell Fragmentationn in the trap is obtained by resonantly accelerating the ions to induce multiplee low-energy collisions with the surrounding helium gas. When the resonancee excitation voltage is too small no fragmentation takes place, and the MS/MSS spectrum is identical to the MS spectrum. With increasing excitation-voltagee fragmentation occurs, first marginally, i.e. protonated molecular ions are stilll dominant, and then prevalently, i.e. protonated molecular ions are not observedd anymore. The total amount of internal energy imparted during CID determiness which fragmentation pathways will be taken. The composition of MS/MSS spectra is therefore depending on the intensity of the controlled collisional-activation.. If the excitation is too small, collisions fail to produce sufficientt diagnostic fragment ions. Conversely, an excessive excitation hinders the formationn of diagnostic fragment ions and the MS/MS spectrum is useless. Correct CIDD intensity must be sought for each particular compound (see section 4.5.4). As

forr single MS experiments, exact reproducibility of the MS/MS sequence of events iss not expected. Distribution and relative intensities of fragment ions slightly vary fromm shot to shot and it is wise to average the spectrum over several shots.

Thee MS/MS experiment can be reiterated several times to investigate the fragmentationn of the molecular ion. Fragment ions are then further investigated in

ann additional CID experiment (MS"). (Section 6.3.3 discusses an MS4 experiment

withh indigo). MS/MS can be used for structural studies of organic pigments in the presencee of a complex matrix or medium. To optimise the structural information acquiredd in MS/MS two approaches are possible. In the first one, the CID amplitudee is optimised to obtain a good balance in the MS/MS spectrum compositionn between molecular ions and diagnostic fragment relative intensities. Inn a second approach, a complementary set of spectra at different CID amplitudes cann be produced and the sum of this information is used for interpretation. Quantitativee determination seems excluded.

3.8.3.8. Conclusion

Inn LDMS analysis, structural information can be optimised by correct preparationn of the sample and tuning of the experimental parameters. New probes weree designed specifically to accommodate various forms of samples within the twoo mass spectrometers Deposition of the sample as a thin film on a metallic supportt or a TLC plate coated with cellulose provides good ionisation yields. In the casee of paint cross-section, great care should be given to obtain surfaces as smooth ass possible. Tuning the laser power density near to slightly above the laser desorptionn and ionisation threshold value offers the best conditions for the formationn of intact molecular ions and limited fragmentation. At the same time, loww laser power density guarantees smallest deterioration of the sample. Converselyy high laser power should be avoided since excessive molecular fragmentationn makes the interpretation of the spectra more difficult. Experimental resultss have shown that optimal desorption and ionisation conditions are obtained inn experiments where multiple laser shots are used in close succession. In LDMS experimentss performed with the ITMS, ion introduction times and mass cut off mustt be tuned to optimise the spectral information. Best multi-stage MS experimentss are obtained by optimising the density of the ion population of the precursorr ions trapped in the cell and then tuning the CID condition to achieve satisfactoryy fragmentation of the parent ions.